Marco J. Russel and Peter Goldsmith
Artificial Neural Networks, Haptics, Magnetic Localization, Pattern Recognition, Position Tracking
This paper presents a novel use of offline-trained neural networks to track a permanent magnet for a haptic feedback system. A grid of 3-axis magnetic field sensors measures the field at several different points. These values are then passed to a Multi-Layer Perceptron (MLP) neural network, which uses the field measurements at each of the sensors to determine the position of the magnet. The neural network is calibrated using 169 calibration points. A number of networks are trained separately, so that the general solving ability of this method can be assessed. The resulting neural network has comparable accuracy to traditional method of magnetic position tracking, but operates much faster and can track the magnet even when its field is distorted by interfering ferrous material or another magnet.
Important Links:
Go Back
